test/normalize-gnumeric

   1 #!/usr/bin/perl -w
   2 # -----------------------------------------------------------------------------
   3
   4 # This script tries to normalize gnumeric files, i.e., remove insignificant
   5 # differences due to versions, environment, or hash ordering.
   6 #
   7 # It is a perl script mucking with an xml file.  Think it over.  It is not
   8 # hard to cheat this, but for purposes of testing code, it is fine.
   9
  10 my @items;
  11 my $item = '';
  12
  13 while (<STDIN>) {
  14     # "x" out version numbers.
  15     if (m|^\s*<gnm:Version\s+Epoch="\d+"\s+Major="\d+"\s+Minor="\d+"\s+Full="[.0-9]+"/>\s*$|) {
  16         s/="[.0-9]+"/="x"/g;
  17     }
  18     if (m|^\s*<office:document-meta |) {
  19         s/( office:version=")[.0-9]+(")/$1XXX$2/;
  20     }
  21
  22     s{(\bxmlns:gnm="http://www.gnumeric.org/v)\d+(\.dtd)"}{$1XX$2};
  23
  24     if (m|^\s*<gnm:PrintInformation>\s*$| .. m|^\s*</gnm:PrintInformation>\s*$|) {
  25         # "x" out margins.
  26         if (m|^\s*<gnm:Margins>\s*$| .. m|^\s*</gnm:Margins>\s*$|) {
  27             s/="[-.0-9a-zA-Z]+"/="x"/g;
  28         }
  29
  30         # "x" out information from cups.
  31         s{(<gnm:(paper|orientation)>).*(</gnm:\2>)}{$1xxx$3};
  32     }
  33
  34     if (m|^\s*<office:meta>\s*$| .. m|^\s*</office:meta>\s*$|) {
  35         s|(<meta:creation-date>).*(</meta:creation-date>)|$1XXX$2|;
  36     }
  37
  38     if (m|^\s*<gnm:name>WorkbookView::do_auto_completion</gnm:name>\s*$| ..
  39         m|^\s*</gnm:Attribute>\s*$|) {
  40         s|(<gnm:value>).*(</gnm:value>)|$1xxx$2|;
  41     }
  42
  43     # Sort names.
  44     if (0 && m|^\s*<gnm:Sheets>\s*$| .. m|^\s*</gnm:Sheets>\s*$|) {
  45         if (m|^\s*<gnm:Names>\s*$| .. m|^\s*</gnm:Names>\s*$|) {
  46             if (m|^\s*<gnm:Names>\s*$|) {
  47                 # Zip.
  48             } elsif (m|^\s*</gnm:Names>\s*$|) {
  49                 print sort @items;
  50                 @items = ();
  51             } else {
  52                 $item .= $_;
  53
  54                 if (m|^\s*</gnm:Name>\s*$|) {
  55                     push @items, $item;
  56                     $item = '';
  57                 }
  58                 next;
  59             }
  60         }
  61     }
  62
  63     print;
  64 }